Research ArticleFlanker Task-Elicited Event-Related Potential Sources ReflectHuman Recombinant Erythropoietin Differential Effects onParkinson’s Patients

Maria L. Bringas Vega ,1,2 Shengnan Liu,1 Min Zhang,1 Ivonne Pedroso Ibañez,2

Lilia M. Morales Chacon,2 Lidice Galan Garcia,3 Vanessa Perez Bocourt,4

Marjan Jahanshahi,1,5 and Pedro A. Valdes-Sosa 1,3

1 e Clinical Hospital of Chengdu Brain Science Institute, MOE Key Lab for Neuroinformation,University of Electronic Science and Technology of China, Chengdu, China2Centro Internacional de Restauracion Neurologica CIREN, La Habana, Cuba3Centro de Neurociencias de Cuba CNEURO, La Habana, Cuba4Miami Dade College, Miami, FL, USA5UCL Queen Square Institute of Neurology, London, UK

Correspondence should be addressed to Maria L. Bringas Vega; maria.bringas@neuroinformatics-collaboratory.org and PedroA. Valdes-Sosa; pedro.valdes@neuroinformatics-collaboratory.org

Received 5 December 2019; Revised 21 April 2020; Accepted 7 May 2020; Published 22 May 2020

Academic Editor: Helio Teive

Copyright © 2020Maria L. Bringas Vega et al. /is is an open access article distributed under the Creative Commons AttributionLicense, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work isproperly cited.

We used EEG source analysis to identify which cortical areas were involved in the automatic and controlled processes of inhibitorycontrol on a flanker task and compared the potential efficacy of recombinant-human erythropoietin (rHuEPO) on the per-formance of Parkinson’s Disease patients./e samples were 18medicated PD patients (nine of them received rHuEPO in additionto their usual anti-PD medication through random allocation and the other nine patients were on their regular anti-PDmedication only) and 9 age and education-matched healthy controls (HCs) who completed the flanker task with simultaneousEEG recordings. N1 and N2 event-related potential (ERP) components were identified and a low resolution tomography(LORETA) inverse solution was employed to localize the neural generators. Reaction times and errors were increased for theincongruent flankers for PD patients compared to controls. EEG source analysis identified an effect of rHuEPO on the lingual gyrifor the early N1 component. N2-related sources in middle cingulate and precuneus were associated with the inhibition ofautomatic responses evoked by incongruent stimuli differentiated PD and HCs. From our results rHuEPO seems to mediate aneffect on N1 sources in lingual gyri but not on behavioural performance. N2-related sources in middle cingulate and precuneuswere evoked by incongruent stimuli differentiated PD and HCs.

1. Introduction

Discovering neuroprotective agents to slow down the pro-gression of Parkinson’s Disease (PD) and, importantly, toimprove cognitive deficits is an active area of research [1]./e search for agents to supplement usual dopaminergictreatments directed towards motor symptoms is not sur-prising since the characteristic motor impairment of patients

is usually accompanied by cognitive deficits [2]. Sincecognitive dysfunction has a negative impact on the quality oflife of patients [3]; finding effective therapies that targetcognition in PD is of paramount importance. As an example,we found that human recombinant erythropoietin(rHuEPO) [4] improved general measures of cognition inchronically medicated PD patients, an additional benefit tothat obtained on their usual medical treatment. /is result

HindawiParkinson’s DiseaseVolume 2020, Article ID 8625794, 10 pageshttps://doi.org/10.1155/2020/8625794

extends to PD the evidence for neuroprotective properties ofrHuEPO already described in other neurologic diseases [5]and is supported by the antiapoptotic, anti-inflammatory,and cytoprotective effects of EPO in PD animal models[6, 7]. /is promising result suggested the need to furtherstudy the effect of rHuEPO on cognition in PD.

We believe that, to further understand the effect ofrHuEPO on cognition in PD patients, we need to examine itseffect on specific stages of information processing. /is isbecause the overt behavioural measures used in our previousstudy (a) do not have temporal sensitivity, being the endoutcome of many sequential processes, and (b) do not reflectlocalized neural activity. Consequently, and as a first ob-jective, we zeroed in on very early automatic neural pro-cesses involved in inhibitory control, the lack of which is socommon in nondemented PD patients. /is early lack ofinhibitory control is easily measured in a number of taskssuch as the Stop signal, go no-go, Stroop, Hayling SentenceCompletion task, and the Simon task described in [8, 9].However, we decided to use a very well-studied paradigm:Ericksen’s Flanker Task [10]. It explores the lack of inhi-bition related to the difficulty in suppressing interference byincongruent stimuli. It allows the evaluation of very shortlatency automatic activation to incongruent flankers around100msec. and other controlled processes around 200msec./ese produce increased reaction times (RTs) and errors inincongruent trials versus congruent trials in PD patients incomparison with normal (e.g., [11–13]. It is, however, theearly ERP responses that are of interest here, not the overtbehavioural response indexed by the RT which occurs laterabout 400msec.

/ere is no clear way to study these early responsesbehaviourally. However, these processes might be probed bydirect measurements of fast neural responses such as thoseprovided by event-related responses (ERPs). In particular,the flanker task elicits the N1, N2, and P3 ERP components,which are related to automatic and controlled process, re-spectively [14]. Here, we will focus only on the earlycomponents N1 and N2./eN1 component has not been, toour knowledge, sufficiently studied in the flanker task in PD.However, the frontocentral N2 on incongruent trials offlanker tasks in patients with PD have received more at-tention [13, 15–18]. /e comparison of medicated PD pa-tients and drug-naıve de novo PD patients showed thatneither the presence of PD (see also [17]) nor dopaminergicmedication modulates N2 amplitude variability on incon-gruent conditions of flanker tasks (for a discussion see areview of ERP and cognition in PD by Seer et al. [19]). Itseems logical then to determine if the additional cognitiveimprovement, produced by rHuEPO with respect to do-paminergic treatment, is accompanied by changes in theearly components in the N1 and N2 ERP components,helping us to pinpoint one of the stages of cognitive pro-cessing affected by this drug. Furthermore, in addition tofiner grained timing information, it is possible to leveragesource localization methods to identify the neural sources ofany ERP component change.

/erefore, the aim of our study is to use a flanker task toidentify if rHuEPO improves automatic and controlled

inhibitory control in PD patients and to locate the neuralgenerators of these processes. /is could be a first step inidentifying an ERP biomarker for this type of cognitiveprocess to be used in clinical trials.

2. Materials and Methods

2.1. Methods

2.1.1. Description of the Sample and Clinical Trial.Eighteen PD patients (Hoehn and Yahr stages I to III, meanage 53.9, SD 3.2 years) were recruited at the Clinic ofMovement Disorders and Neurodegeneration, Centro In-ternational de Restauracion Neurologica (CIREN) in LaHabana, Cuba, to participate in a safety clinical assay ofErythropoietin (rHuEPO) in PD. /e design of this inves-tigation, results, scheme of application, and doses employedmay be found in [4]. Inclusion criteria were a clinical di-agnosis of idiopathic PD according to the UK Brain Bankcriteria and a good response to dopaminergic treatment andaged between 45 and 75 years [20]. Exclusion criteria weremanifestation or indicative signs of major cognitive im-pairment, psychotic symptoms, and/or presence of otherchronic diseases. Nine of the PD patients, through randomallocation, received additionally to their usual anti-Parkin-sonism medication rHuEPO for five weeks and the othernine did not. rHuEPO approved and registered for use inhumans was obtained at the Centro de Inmunologia Mo-lecular, La Habana Cuba (ior® EPOCIM). /ere were nosignificant differences in age, years of education, or durationof illness between the two PD groups. To exclude dementiaand major depression, the Mini-Mental State Examinationand the Hamilton Depression Scale were, respectively, ad-ministered [21, 22]. All patients were assessed on the motorsubscale of the Unified Parkinson´s Disease Rating Scale(UPDRS) both during “on” (mean 6.3, SD1.1) and “off”medication (mean 21.7, SD 4.3) states.

For the purpose of comparisons, 9 healthy controls(HCs) matched in age (mean 51.2, SD 3.9 years) and edu-cational level were recruited at the same clinic. /e PDpatients were tested on their usual anti-Parkinsonismmedication. /e patients and controls signed an informedconsent to participate in this study as a complement of theclinical trial following the CIREN ethics committeeregulations.

2.1.2. Eriksen’s Flanker Task. All participants completed theEriksen’s Flanker Task, while the EEG was simultaneouslyrecorded. Each trial of the task consisted of the presentationof a set of 5 ordered letters (HHHHH or SSSSS) for thecongruent condition and 5 letters with H or S at the centreand different laterals or flankers (SSHSS or HHSHH) for theincongruent condition. Participants were instructed to re-spond to the central letter, whether H or S, by pressing a keywith the index finger of the right or left hand, respectively.Participants were instructed to respond as fast and as ac-curately as possible. A total of 480 trials in two blocks, eachlasting 8 minutes, were completed. In each block 80 stimuliwere shown for the congruent condition and 160 for the

2 Parkinson’s Disease

incongruent. Only the correct responses with reaction times(RTs) >150 and <800msec were selected for analysis.

/e physical characteristics of the stimuli were blackletters on a white frame with a height� 1.5 cm andlength� 7 cm, under 6° visual angle. /e distance of theparticipant to the computer monitor was 60 cm. Eachstimulus was presented at the centre of the screen for190msec, followed by a fixed interstimulus interval (ITI) of1735msec. A training block of 40 stimuli was designed toensure task instructions were understood.

2.1.3. ERP Measurement. /e electroencephalogram (EEG)was continuously recorded at a sampling rate of 512Hz from64 electrodes located at standard positions of the Interna-tional 10/20 System using a Brain Vision system (https://www.brainproducts.com/products_by_apps.php?aid�5)[23]. Linked ears were used as online reference and the frontas Earth. To monitor eye movement artefacts, the electro-oculogram (EOG, horizontal and vertical) was recordedfrom electrodes placed 1 cm to the left and right of theexternal canthi, and from an electrode beneath the right eye.

Data were filtered using 1–30Hz and a notch filter toeliminate the 60Hz powerline artefact. All data were ref-erenced using an average reference to all the channels. /ebaseline was corrected between −200 and 0msec. Epochswith electric activity exceeding baseline activity by 100 μVwere considered as artefacts and were automatically rejectedfrom further processing (15% of epochs related to hits and11% of the epochs related to errors). For the analysis, severalelectrodes were excluded (EOG, ECG, TP9, and TP10).

ERPs were obtained from the EEG recordings for eachparticipant for all the electrodes within the two experimentalconditions and averaged over the two groups using Analyzersoftware (https://www.brainproducts.com/productdetails.php?id�17). Epochs of 800msec (from −200msec (base-line) until 600msec poststimulus onset) were analyzedlocked to the stimulus. We selected two windows to examinethe stimulus-locked ERPs, using only the correct responseaverages for the N1 (80–180msec.) and N2 (200–300msec.)components in the expected time windows (see ERPsguidelines in [24]. Henceforth we will refer to these averagessimply as the amplitude of the N1 and N2 components. /eaverage waveform for each participant and each conditionwas estimated in all the electrodes, but the averagedwaveform for group is plotted below for the electrode withthe higher statistics amplitudes.

In order to localize the generators of the ERP compo-nents, a lead field was constructed for each participant tocalculate the (volume-constrained) inverse solution, at thetwo selected latencies using LORETA (low resolution to-mography) (http://www.uzh.ch/keyinst/loreta) [25]. ForLORETA, the intracerebral volume is partitioned into 6239voxels at 5mm spatial resolution.

2.2. Statistical Analysis. We now summarize the experi-mental design. Our sample is divided into 3 groups: 9Parkinson patients with the usual treatment (PD Control), 9patients with the usual treatment plus EPO (PD rHuEPO),

and 9 healthy controls (HCs). Additionally, the ERPs foreach participant were recorded in two conditions: congruentand incongruent.

For each participant the following variables were used inthis paper:

(1) Reaction time and errors to the Flanker task(2) Amplitude of the N1 and N2 ERP component at the

60 EEG scalp electrodes(3) Power of the N1 and N2 sources component for the

6239 source voxels

/e statistical analyses performed were as follows:

(a) Reaction times and errors were analyzed using a two-way repeated measure ANOVA with the group(HCs, PDControl, and PD rHuEPO) as the between-group factor and the experimental condition (in-congruent versus congruent) as the within-subjectrepeated measures factor. We report the F statisticand the p value for tests of the main effect and theinteraction. /e Greenhouse-Geisser adjustmentwas applied since lack of sphericity was observed./ese analyses were completed with STATISTICA7.0.

(b) An exploratory analysis of the differences in ERPamplitude topographies between the HCs and PDControl + PD rHuEPO groups was carried out bymeans of a multivariate t-test that corrects formultiple comparisons by means of a permutationtechnique. /e permutation test has the followingadvantages: the tests are distribution-free that con-trol the experimentwise error for the simultaneousunivariate comparisons, no assumptions of an un-derlying correlation structure are required, and theyprovide exact p values valid for any number ofsubjects, timepoints, and all 60 electrodes. /eoverall significance level was selected to be 0.05. /emethod is described in [26, 27] as implemented inthe software NEEST from Neuronic http://www.neuronicsa.com/. /is allowed the selection of thefollowing:

(1) A subset of electrodes to be subjected to Mul-tivariate Analysis of Variance (MANOVA)(described in (c))

(2) /e selection of most representative electrodes toplot the N1 and N2 grand average ERPs

(3) /e analysis of time intervals to be furtherstudied

(c) Examine for each ERP component and for theirselected group of electrodes repeated measuresMultivariate Analysis of Variance (r-MANOVA) forthe design Group by Condition with a significancelevel set at the 0.05 level. /e different contrasts forthe interaction and main effects were tested by usingWilk’s lambda, approximated by an F function andthe p value reported. Note that this allows a

Parkinson’s Disease 3

simultaneous confidence interval for contrasts ongroup differences and to examine which electrodecontributes to the effects. /e MANOVA was thatimplemented in the STATISTICA 7.0. package.

(d) Further analysis for selected differences of the ERPcomponent source images between selected groupswas carried out using the LORETA-built-in voxel-wise randomization tests with 2000 permutations[28], based on statistical nonparametric mapping.Voxels with significant differences (p< 0.01, cor-rected for multiple comparisons) between contrastedconditions were located with the coordinates of theAAL (Automated Anatomical Labelling of Activa-tions) 116 structures Atlas of the Montreal Neuro-logical Institute (MNI) [29].

3. Results

3.1. Behavioural Results

3.1.1. Reaction Time. /e differences between the threegroups were significant for factor group (F(2, 24)� 7.47,p � 0.003), the Condition was not significant as we predictedin the preliminary analysis (F(2, 24)� 3.22, p � 0.06). /einteraction of Group∗Condition also was not significant(p> 0.8). /e contrast between the two groups of patients(PD Control and PD rHuEPO) did not show differences inthe reaction time (F(2, 15)� 0.62, p � 0.55). Table 1 showsthe performance of the PD groups separately and Table 2 thefusion of PD patients versus HCs.

3.1.2. Errors. /e differences between the errors in the threegroups were significant for Factor Group (F(2, 15)� 10.49,p � 0.0014) and for Condition (F(2, 24)� 11.6, p � 0.0003),but not for the interaction Group∗Condition (p � 0.1). /ecomparison between the two PD groups was significant onlyfor Condition, incongruent (F(1, 16)� 55.3, p � 0.00001,and not for the congruent condition (F(1, 16)� 1.88,p � 0.18).

When using the contrast comparing all PD patients andHCs (Table 2), the results were consistent with previousfindings where the RTs increased with incongruent flankerscompared to congruent for both groups.

3.2. Exploratory Results of ERPs. As mentioned in Section2.1, the multivariate t tests corrected for multiple compar-isons with permutation tests provides exact p values, validfor any number of participants, timepoints, and recordingsites yielded as significant the ERP components in themidline at the 0.05 level. Within this group, the most sig-nificant ERP was Oz for N1 and Cz for N2 as described in theliterature. We will therefore concentrate on these electrodesets henceforth since they all are significant above theglobally valid significance threshold.

/e same procedure allows, additionally, to select thetime windows and which factor (Condition or Group) to befurther analyzed. Figure 1 illustrates, for one derivation, thestatistics shown above the red line, the latencies with

significance for each factor (Group or Condition) in all thetime window for analysis. /e interaction between them wasnot significant at any time./e exploratory analysis betweenexperimental conditions did not reflect significant differ-ences in the time range for the early ERP components N1and N2 (around 100 and 200msec, respectively).

Note that the significant differences for Condition are inthe range of the P300 or later, not in the scope of our study.For that reason, we focus all the further analysis on theincongruent condition, which is the condition which elicitsinhibitory control. Nevertheless, henceforth we continue toreport the full two-way analysis (Group×Condition),though concentrating on the Group Factor analyses.

3.2.1. Analysis of the N1 Component. We tested the N1amplitudes with the repeated measures rMANOVA(Group×Condition) and examined the main effects and theinteraction between them. /e interaction and the factorCondition were not significant (p � 0.23). However, themain effect of Group was significant with Wilk’sLambda� 0.40, F(8, 42)� 2.97, p � 0.009. A contrast be-tween the two groups of patients was also significant withWilk’s Lambda� 0.47, F(4, 13)� 1.2, p � 0.003. Further-more, with electrodewise contrasts 13 electrode sites F4,FC2, FC4, FC6, C2, C4, C6, CP2, O1, O2, Oz, PO3, PO4,PO7, PO8 retained significance. Note that the N1 at the O1electrode followed the following pattern (see Figure 2): theamplitude of the PD rHuEPO group (−4.2 μV) was notdifferent statistically from that of the HCs. On the otherhand, the amplitude of the PD Control group (−1.2 μV) wassignificantly lower.

/e localization of the differences between the twoParkinson groups of this component is localized anatomi-cally by means of the randomized nonparametric test forLORETA./is showed that the PD rHuEPO had a larger N1component than the PD Control group at the p< 0.001 level(corrected for multiple comparisons) at the lingual gyri (seeFigure 2(b)).

3.2.2. Analysis of the N2 Component. We tested the N2amplitudes with the repeated measures rMANOVA(Group×Condition) and examined the interaction and themain effects. /e interaction was not significant with Wilk’sLambda� 0.43, F(6, 44)� 2.97 and the factor Condition wasalso not significant (p � 0.323).

/e main effect of Group (comparing three groups) wassignificant, F(2, 24)� 6.14, p � 0.006, in seven frontocentralelectrodes: Cz (F(2, 24)� 6.50, p � 0.005), CPz (F(2, 24)�

4.43, p � 0.02), CP1 (F(2, 24)� 5.9, p � 0.008), CP2 (F(2,24)� 5.6966, p � 0.00945), C1(F(2, 24)� 3.6125,p � 0.04251), C2 (F(2, 24)� 4.6242, p � 0.02).

A contrast between the two groups of patients was alsosignificant in frontocentral areas, the electrodes Cz (F(2,24)� 4.43, p � 0.002), CPz (F(2, 24)� 6.5, p � 0.005) andFC1, FC2, C1, C2 (p< 0.05). /ere were no significantdifferences between conditions or interaction betweenfactors.

4 Parkinson’s Disease

CZ electrode

Latencies in milliseconds

T-te

st va

lues

15

10

5

0

25

20

15

10

5

00.00 50.00 116.00 188.00 262.00 334.00 406.00 480.00 50.000.00 116.00 188.00 262.00 334.00 406.00 480.00

Figure 1: Left: t values for the tests of differences between Groups independent of Condition. Right: the t values’ tests for differences betweenCondition independent of Group. /e red line indicates the statistical significant threshold (corrected for all electrodes and all times by amultivariate permutation test).

–200 0 200 400 600 800 1000 1200–2.5

–2

–1.5

–1

–0.5

0

0.5

1

1.5

Am

plitu

de (μ

V)

Time (ms)

Healthy controlPD controlPD rHUEPO

(a)

50.00000

9.44060

(b)

Figure 2: (a) /e group average N1 waveform for each group in the window (80–180msec) in the electrode site O1 with the highestamplitude. /e N1 peak was at 152msec. (b) /e lingual gyri are the sources of the N1 component according to AAL coordinates (X� 92,Y� 76, and Z� 172). /e scale of statistical significance is self-generated using the real values of the original data. All the voxels plotted weresignificant at p< 0.01.

Table 1: /e results of the reaction times and the percent errors for the congruent and incongruent trials for the PD patients with andwithout rHuEPO. /e values in the table are means with standard deviations in parenthesis.

PD rHuEPO n� 9 PD Control n� 9Congruent Incongruent Congruent IncongruentMeans (SD) Means (SD) Means (SD) Means (SD)

Reaction times (msec) 459.33 (71.76) 479.89 (49.43) 460.22 (72.10) 488.22 (63.76)Percent errors 13.22 (7.76) 43.22 (21.37) 8.78 (6.76) 32.00 (15.79)

Table 2: /e results of the reaction times and percent errors for the congruent and incongruent trials for the Parkinson’s. disease (PD)patients and healthy control (HCs) groups. /e values in the table are means with standard deviations in parenthesis.

(PD rHuEPO+PD Control) n� 18 HCs n� 9Congruent Incongruent Congruent IncongruentMeans (SD) Means (SD) Means (SD) Means (SD)

Reaction time (msec.) 459.78 (69.79) 484.06 (55.51) 411.22 (52.00) 431.33 (43.47)Percent errors 9.00 (3.81) 37.61 (19.12) 3.33 (2.40) 11.00 (7.42)

Parkinson’s Disease 5

Note that the N2 grand average at the Cz electrode fol-lowed an opposite pattern compared to N1 (see Figure 3(a)):the amplitude of the PD Control group (−2.10 μV) andhealthy controls (−2.46 μV) was not different statistically. Onthe other hand, the amplitude of the PD rHuEPO group(−0.67 μV) was significantly lower than both of them. SeeTable 3 for details of amplitude and latencies of N2 in Cz.

/e source analysis of the differences (comparing thethree groups), for the N2 component, was localized ana-tomically by means of the LORETA randomized nonpara-metric test (p< 0.01 level corrected for multiplecomparisons) at the middle cingulum and precuneus bi-laterally. See Figure 3(b).

In order to know if the errors were related to the N2amplitude, we select a linear mixed effect model and carriedout a repeated measures ANOVA log(errors) ×Group×N2amplitude. But the results were not significant for the in-teraction of log(errors) with the N2 amplitude, only themaineffect for Group (p � 0.001675) (see Figure 4).

4. Discussion

/e current study was designed to examine if the novelrHuEPO neuroprotective compound, given to Parkinsonpatients in addition to their usual medication, changed theamplitude of ERP components during an inhibitory controltask.

/e behavioural results were consisted with previousstudies in PD patients in both the rHuEPO and PD Controlgroups. Both groups showed significantly increased reactiontimes and a higher number of errors to the incongruentstimuli during the performance of the flanker task ascompared to age and education-matched HCs. /ese highererror rates in PD Controls are consistent with the proposalthat the basal ganglia together with the anterior cingulate[30] participate in the monitoring of incongruence and errormonitoring [31, 32] which may be impaired in PD due to thedopamine deficiency (for a recent revision of how theprogressive dopamine deficiency reduces striatal cholinergicinterneuron activity; see [33].

It should be noted that we did not find the expectedbeneficial effect of rHuEPO on behavioural performance(RT and accuracy) in PD patients who received the neu-roprotective agent as compared to those that only receivedthe usual treatment. Rather, the differences between groupsof patients were found in the ERP components. /is is inaccordance with our hypothesis that an overall behaviouralresponse might be noisier than some of its time parsedsubstages. /is suggests further studies to identify overtbehavioural responses at similar short time scales as ERPcomponents. On the other hand, as it sometimes happenswith this type of clinical study the small sample size may leadto lack of power to detect subtle effects.

4.1. Regarding the N1 Component. /is component reflectsselective attention, linked to the basic characteristics of astimulus, and also to the recognition of a specific visualpattern [34]. N1 amplitude also has been hypothesized to

reflect sustained covert visual attention [35] being associatedwith the intensity of covert attention to the central target inthe flanker task. In terms of spatial localization, the N1amplitude is greater in occipital regions [34, 36]. /e neuralsources of the N1 in flanker tasks were located at the brainvisual areas of the occipital cortex [34, 37, 38]. For example,Bokura et al. [39] using LORETA identified additionalsources of the visual N1 in the occipitotemporal lobe [39]and Zhang et al. [40] also localized N1 for flanker source inextrastriate visual cortex. We thus expected the differencesbetween PD groups to be localized on the scalp in the oc-cipital electrodes and the sources to be in brain occipitalareas.

/is is what we found: the generators of N1, both in thescalp topography and using LORETA, in the visual areas ofthe occipital lobe of both hemispheres. /e activation of thesource for the PD patients who received rHuEPO was muchlarger than that of the PD group who did not receive it. Infact, the response of the rHuEPO group became statisticallyindistinguishable from that of the HCs, suggestive of apossible neuroprotective effect of rHuEPO on the lingualgyrus, a region associated with the early and automaticprocessing of visual stimuli. In summary, our findingssuggested an effect of rHuEPO on the visual attentionalwindow in the early information-processing stage, thusenhancing the automatic processing of flankers regardless oftheir compatibility.

4.2. Regarding the N2 Component. /e second componentN2 has been found in several studies of inhibition using theFlanker task and its amplitude and latency was unaltered inmedicated PD patients (for review, see [9]). Veen and Carter[41] used BESA source localization to study inhibition andresponse conflict in the Eriksen’s Flanker Task, determiningthat the N2 amplitude associated with incongruent trials canbe explained by a dipole that is located in the ACC. Bokuraet al. [39] also conducted an experiment to understand theanatomical structures that are involved in N2 using a visualmodality of the flanker paradigm and LORETA which lo-cated the N2 generators at cingulate and the right lateralorbitofrontal cortex.

In our study, we found that the amplitude of N2component for the PD Control and HC groups was sta-tistically indistinguishable. But the N2 amplitude in therHuEPO PD group was diminished with respect to the othertwo groups. /ese effects were topographically located, asexpected, in the frontocentral areas, with neural generatorsof these differences localized to the posteromedial portion ofthe parietal lobe, the precuneus, a structure involved in theprocessing of perceptual ambiguities of stimuli [42], and inthe middle cingulate cortex, probably related to monitoringof conflict in the Flanker task [43]. In comparison withprevious reports, we concur with Van Eimeren who founddysfunction of the default mode network and particularlydeactivation of the posterior cingulate cortex and the pre-cuneus [44] in PD relative to healthy controls, consideringthese changes in PD closely related to higher errors in ex-ecutive tasks in PD compared with healthy controls.

6 Parkinson’s Disease

Time (ms)–300 –200 –100 0 100 200 300 400 500 600

–3

–2

–1

0

1

2

3

Am

plitu

de (μ

V)

Healthy controlPD controlPD rHUEPO

(a)

30.30000

9.32030

(b)

Figure 3: (a)/e N2 waveform averaged by groups in the window (200–300msec) in the electrode site Cz with the highest amplitude. Notethat, for the HC group, the early 195msec latency and for both PD patients a later peak around 224msec. (b) /e N2 component showedmaximal activation at middle cingulum and precuneus bilaterally (left located at X� 92, Y� 108, Z� 156). To the right, the localization of theprecuneus left. /e bicolour scale is showing all the significant values after Bonferroni correction and using permutations.

Table 3:/emeasures of amplitude and latency of the N2 component for the two conditions congruent and incongruent at the electrode Czwhich exhibited the highest amplitude.

GruposAmplitude (μV) condition Latency (msec.) condition

Cz-cong Cz-Incong Cz-cong Cz-IncongHCs −2.4 − 2.45 195 199PD cCControl − 2.10 − 2.09 226 224PD rHuEPO − 0.56 − 0.67 224 223

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

–5 –4 –3 –2 –1 0 1 2 3

Log

erro

rs

Amplitude N2

PD rhuEPOPD controlHC

Figure 4:/e plot of the N2 and log(errors) of the three groups. Note the variability of the data with 2 outliers of the HCs and 1 outlier of thePD Control group with positive amplitudes of N2.

Parkinson’s Disease 7

However, in our study, the striking decrease of the N2produced by rHuEPO needs further research to find anadequate explanation.

4.3. Behaviour versus ERPs. Contrary to our expectation,rHuEPO was not associated with a significant improvementin behavioural performance and did not influence the neuralgenerators of the N2.

/e ERP allows neural activity tracking on a millisecondtime scale and represents a continuous measure of infor-mation processing; for this reason, we selected the ERP tostudy a more refined measure of the process of inhibitorycontrol.

/is apparent contradiction between behavioural andelectrophysiological results could be related to their differenttemporal course. Note that the inhibition is a complexprocess that can be automatically initiated in the first100msec after stimulus and extend its action through bothautomatic and controlled processes until 800msec. Reactiontime, on the other hand, started much later >400 millisec-onds after the stimulus presentation, with a strong motorcomponent to complete the response.

/erefore, the aim of our study is to use a flanker task toidentify if rHuEPO improves automatic and controlledinhibitory control in PD patients and to locate the neuralgenerators in these processes. /is could be a first step inidentifying an ERP biomarker for this type of cognitiveprocess to be used in clinical trials.

4.4. Limitations. Since this study was completed as part of asafety trial, the samples and the doses employed were small./is might also explain the lack of clear correlations withbehaviour, for example, reaction time with N2 amplitude./us, the results require confirmation with larger samples infuture studies. However, the results highlighted the role ofEEG source analysis and advantages of electrophysiologywith its high temporal resolution and insensitivity to placeboeffects, in identifying brain changes after an interventionsuch as rHuEPO.

5. Conclusions

(i) We found that rHuEPO improved automatic in-hibitory control in PD patients but did not improvebehavioural performance.

(ii) /e differences between PD rHuEPO and PDControl groups were in the N1 component at thelingual gyrus. /e differences between PD andhealthy controls were on the N2 component in thecingulate and precuneus.

(iii) Electrophysiology is potentially a useful tool foridentifying effects of neuroprotective compoundson different stages of information processing.

(iv) /e components N1 and N2 as well as others like P3should be further studied as possible biomarkers forthe evaluation of neuroprotective drugs in Par-kinson’s disease.

Data Availability

/e tables with the behavioural performance (reaction time,hits, and errors) and the N2 amplitude for the averaged timewindow (200–300 milliseconds) of the samples were sub-mitted in the supplementary material 1. /e raw and pre-processing EEG recordings in Brain Vision format with allthe individual and grand average potentials for group andcondition can be available upon request to maria.bringas@neuroinformatics-collaboratory.org.

Conflicts of Interest

/e authors declare that there are no conflicts of interestregarding the publication of this paper.

Authors’ Contributions

Maria L. Bringas Vega, and Pedro A. Valdes-Sosa con-tributed equally to this paper.

Acknowledgments

/is paper received support from the NSFC (China-Cuba-Canada) project (no. 81861128001) and the funds fromNational Nature and Science Foundation of China (NSFC),with funding nos. 61871105, 61673090, and 81330032, andCNS Program of UESTC (no. Y0301902610100201). /eauthors would like to thank to the Centro de Neurocienciasde Cuba, specially to Valia Rodriguez and Indira Alvarez fortheir support during the recordings of the ERPs and theCentro Internacional de Restauracion Neurologica for therecruitment and neuropsychological evaluation of the pa-tients. /e authors are in debt with all the PD patients andtheir caretakers who volunteered to participate in our study.

Supplementary Materials

/e number 1 in the initial submission contained the tableswith the raw behavioural performance of the participants inthe flanker task: reaction time, errors, and hits. /e secondsupplementary material we uploaded contained the rawbehavioural performance of the participants in the flankertask: reaction time, errors, and hits. And we added the finaltables of the output of the statistical analysis and the am-plitude of the N2 component. (Supplementary Materials)

References

[1] D. Athauda and T. Foltynie, “/e ongoing pursuit of neu-roprotective therapies in Parkinson disease,” Nature ReviewsNeurology, vol. 11, no. 1, pp. 25–40, 2015.

[2] A. A. Kehagia, R. A. Barker, and T. W. Robbins, “Neuro-psychological and clinical heterogeneity of cognitive im-pairment and dementia in patients with Parkinson’s disease,” e Lancet Neurology, vol. 9, no. 12, pp. 1200–1213, 2010.

[3] A. Schrag, M. Jahanshahi, and N. Quinn, “What contributes toquality of life in patients with Parkinson’s disease?” Journal ofNeurology, Neurosurgery & Psychiatry, vol. 69, no. 3,pp. 308–312, 2000.

8 Parkinson’s Disease

[4] I. Pedroso, M. L. Bringas, A. Aguiar et al., “Use of Cubanrecombinant human erythropoietin in Parkinson ’ s diseasetreatment,” MEDICC Review, vol. 14, no. 1, pp. 11–17, 2012.

[5] M. Brines and A. Cerami, “Emerging biological roles forerythropoietin in the nervous system,” Nature ReviewsNeuroscience, vol. 6, no. 6, pp. 484–494, 2005.

[6] A-L. Siren, T. Faßhauer, C. Bartels, and H. Ehrenreich,“/erapeutic potential of erythropoietin and its structural orfunctional variants in the nervous system,”Neurotherapeutics,vol. 6, no. 1, pp. 108–127, 2009.

[7] Y.-Q. Xue, L.-R. Zhao, W.-P. Guo, and W.-M. Duau,“Intrastriatal administration of erythropoietin protects do-paminergic neurons and improves neurobehavioral outcomein a rat model of Parkinson’s disease,” Neuroscience, vol. 146,no. 3, pp. 1245–1258, 2007.

[8] I. Duan, L. Wilkinson, E. Casabona et al., “Deficits in in-hibitory control and conflict resolution on cognitive andmotor tasks in Parkinson’s disease,” Experimental BrainResearch, vol. 212, no. 3, pp. 371–384, 2011.

[9] C.Macıas, F. Lange, D. Georgiev, M. Jahanshahi, and B. Kopp,“Event-related potentials and cognition in Parkinson’s dis-ease: an integrative review,” Neuroscience & BiobehavioralReviews, vol. 71, pp. 691–714, 2016.

[10] B. A. Eriksen and C. W. Eriksen, “Effects of noise letters uponthe identification of a target letter in a nonsearch task,”Perception & Psychophysics, vol. 16, no. 1, pp. 143–149, 1974.

[11] P. Praamstra, D. F. Stegeman, A. R. Cools, andM. W. Horstink, “Reliance on external cues for movementinitiation in Parkinson’s disease. Evidence from movement-related potentials,” Brain, vol. 121, no. 1, pp. 167–177, 1998.

[12] P. Praamstra, E. M. Plat, A. S. Meyer, andM.W. I. M. Horstink, “Motor cortex activation in Parkinson’sdisease: dissociation of electrocortical and peripheral mea-sures of response generation,” Movement Disorders, vol. 14,no. 5, pp. 790–799, 1999.

[13] S. A. Wylie, J. C. Stout, and T. R. Bashore, “Activation ofconflicting responses in Parkinson’s disease: evidence fordegrading and facilitating effects on response time,” Neuro-psychologia, vol. 43, no. 7, pp. 1033–1043, 2005.

[14] L. Pires, J. Leitão, C. Guerrini, and M. R. Simões, “Event-related brain potentials in the study of inhibition: cognitivecontrol, source localization and age-related modulations,”Neuropsychology Review, vol. 24, no. 4, pp. 461–490, 2014.

[15] M. Falkenstein, R. Willemssen, J. Hohnsbein, andH. Hielscher, “Effects of stimulus-response compatibility inParkinson’s disease: a psychophysiological analysis,” Journalof Neural Transmission, vol. 113, no. 10, pp. 1449–1462, 2006.

[16] J. R. Folstein and C. Van Petten, “Influence of cognitivecontrol and mismatch on the N2 component of the ERP: areview,” Psychophysiology, vol. 45, pp. 152–170, 2008.

[17] R. Verleger, J. Hagenah, M. Weiss et al., “Responsiveness todistracting stimuli, though increased in Parkinson’s disease, isdecreased in asymptomatic PINK1 and Parkin mutationcarriers,” Neuropsychologia, vol. 48, no. 2, pp. 467–476, 2010.

[18] S. A. Klein, W. P. M. van den Wildenberg, K. R. Ridderinkhofet al., “/e effect of Parkinson’s disease on interferencecontrol during action selection,” Neuropsychologia, vol. 47,no. 1, pp. 145–157, 2009.

[19] C. Seer, F. Lange, S. Loens et al., “Dopaminergic modulationof performance monitoring in Parkinson’s disease: an event-related potential study,” Scientific Reports, vol. 7, no. 1,pp. 1–13, 2017.

[20] A. J. Hughes, S. E. Daniel, L. Kilford, and A. J. Lees, “Accuracyof clinical diagnosis of idiopathic Parkinson’s disease: a

clinico-pathological study of 100 cases,” Journal of Neurology,Neurosurgery & Psychiatry, vol. 55, no. 3, pp. 181–184, 1992.

[21] M. F. Folstein, S. E. Folstein, and P. R. McHugh, “Mini-mentalstate,” Journal of Psychiatric Research, vol. 12, no. 3,pp. 189–198, 1975.

[22] M. Hamilton, “A rating scale for depression,” Journal ofNeurology, Neurosurgery & Psychiatry, vol. 23, no. 1,pp. 56–62, 1960.

[23] H. H. Jasper, “Report of the committee on methods of clinicalexamination in electroencephalography,” Electroencephalog-raphy and Clinical Neurophysiology, vol. 10, no. 2, pp. 370–375, 1958.

[24] T. W. Picton, S. Bentin, P. Berg et al., “Guidelines for usinghuman event-related potentials to study cognition: recordingstandards and publication criteria,” Psychophysiology, vol. 37,no. 2, pp. 127–152, 2000.

[25] R. D. Ruchkin, D. Lehmann, T. Koenig et al., “Low resolutionbrain electromagnetic tomography (LORETA) functionalimaging in acute, neuroleptic-naive, first-episode, productiveschizophrenia,” Psychiatry Research: Neuroimaging, vol. 90,no. 3, pp. 169–179, 1999.

[26] L. Galan, R. Biscay, J. L. Rodrıguez, M. C. Perez-Abalo, andR. Rodrıguez, “Testing topographic differences between eventrelated brain potentials by using non-parametric combina-tions of permutation tests,” Electroencephalography andClinical Neurophysiology, vol. 102, no. 3, pp. 240–247, 1997.

[27] L. Galan, R. Biscay, P. Valdes, L. Neira, and T. Virues,“Multivariate statistical brain electromagnetic mapping,”Brain Topography, vol. 7, no. 1, pp. 17–28, 1994.

[28] T. E. Nichols and A. P. Holmes, “Nonparametric permutationtests for functional neuroimaging: a primer with examples,”Human Brain Mapping, vol. 15, no. 1, pp. 1–25, 2001.

[29] N. Tzourio-Mazoyer, B. Landeau, D. Papathanassiou et al.,“Automated anatomical labeling of activations in SPM using amacroscopic anatomical parcellation of the MNI MRI single-subject brain,” Neuroimage, vol. 15, no. 1, pp. 273–289, 2002.

[30] M. M. Botvinick, J. D. Cohen, and C. S. Carter, “Conflictmonitoring and anterior cingulate cortex : an update,” Trendsin Cognitive Sciences, vol. 8, no. 12, pp. 539–546, 2004.

[31] M. Brazdil, R. Roman, M. Falkenstein, P. Daniel, P. Jurak, andI. Rektor, “Error processing-evidence from intracerebral ERPrecordings,” Experimental Brain Research, vol. 146, no. 4,pp. 460–466, 2002.

[32] M. Falkenstein, J. Hoormann, S. Christ, and J. Hohnsbein,“ERP components on reaction errors and their functionalsignificance: a tutorial,” Biological Psychology, vol. 51, no. 2-3,pp. 87–107, 2000.

[33] J. W. Hohnsbein, Z. Shi, I. Kawikova et al., “Dopamine de-ficiency reduces striatal cholinergic interneuron function inmodels of Parkinson’s disease,” Neuron, vol. 103, no. 6,pp. 1056.e6–107, 2019.

[34] S. J. Vahedipour, G. F. Woodman, and E. K. Vogel, “Event-related potential studies of attention,” Trends in CognitiveSciences, vol. 4, no. 11, pp. 432–440, 2000.

[35] F. Di Russo, A. Martinez, and S. A. Hillyard, “Source analysisof event-related cortical activity during visuo-spatial atten-tion,” Cerebral Cortex, vol. 13, no. 5, pp. 486–499, 2003.

[36] G. R. Mangun and S. A. Hillyard, “Allocation of visual at-tention to spatial locations: tradeoff functions for event-re-lated brain potentials and detection performance,” Perception& Psychophysics, vol. 47, no. 6, pp. 532–550, 1990.

[37] C. S. Herrmann and R. T. Knight, “Mechanisms of humanattention: event-related potentials and oscillations,”

Parkinson’s Disease 9

Neuroscience & Biobehavioral Reviews, vol. 25, no. 6,pp. 465–476, 2001.

[38] S. A. Hillyard and L. Anllo-Vento, “Event-related brain po-tentials in the study of visual selective attention,” Proceedingsof the National Academy of Sciences, vol. 95, no. 3, pp. 781–787, 1998.

[39] H. Bokura, S. Yamaguchi, and S. Kobayashi, “Electrophysi-ological correlates for response inhibition in a Go/NoGotask,” Clinical Neurophysiology, vol. 112, no. 12,pp. 2224–2232, 2001.

[40] R. Zhang, M. D. Brandt, W. Schrempf, C. Beste, andA.-K. Stock, “Neurophysiological mechanisms of circadiancognitive control in RLS patients-an EEG source localizationstudy,” NeuroImage: Clinical, vol. 15, pp. 644–652, 2017.

[41] V. v. Veen and C. S. Carter, “/e timing of action-monitoringprocesses in the anterior cingulate cortex,” Journal of Cog-nitive Neuroscience, vol. 14, no. 4, pp. 593–602, 2002.

[42] A. E. Cavanna and M. R. Trimble, “/e precuneus: a review ofits functional anatomy and behavioural correlates,” Brain,vol. 129, no. 3, pp. 564–583, 2006.

[43] S. Enriquez-Geppert, T. Eichele, K. Specht, H. Kugel,C. Pantev, and R. J. Huster, “Functional parcellation of theinferior frontal and midcingulate cortices in a flanker-stop-change paradigm,” Human Brain Mapping, vol. 34, no. 7,pp. 1501–1514, 2013.

[44] T. van Eimeren, O. Monchi, B. Ballanger, and A. P. Strafella,“Dysfunction of the default mode network in Parkinsondisease: a functional magnetic resonance imaging study,”Archives of Neurology, vol. 66, no. 7, pp. 877–883, 2009.

10 Parkinson’s Disease